Take a look at the dashboard dial above. Pretty isn't it? But let's just consider what it's really telling you.

Supposing a mobile communications company wanted to monitor calls connected each second. In a traditional BAM environment, this dial may show the average call/second. Now assume that the operator has more than a thousand cell sites providing coverage to its customers. It's not possible for one person to monitor 1,000 different sites, so the dial shows an aggregate reflecting performance across the whole network over the last 30 minutes. The needle is pointing at 500 calls per second, and it's in the green arc.

But is this is good or not?

The first dial is hiding critical detail. Transactions from one cell site in the capital city are lower than usual, due to one of several air conditioning units not working at full capacity. The cell phone company doesn't know it yet, but the cell site is going to fail completely in the next 24 hours unless a maintenance engineer can get there. A cell site failure can cost in excess of $2m per day in lost revenue.

Unfortunately the dial doesn't show that the volumes at the cell site in trouble are being masked by an adjacent site that is picking up extra volume.

All appears OK.

Event Intelligence can tell you

Using real time event intelligence SeeWhy tackles these types of monitoring scenarios completely differently. By monitoring every cell site uniquely and comparing it with it's individual normal profile, exactly these types of individual problems can be spotted. An alert is automatically sent to network control, highlighting the problem and perhaps presenting a host of sensor metrics which indicate that temperatures in the cell site are up slightly, suggesting the cause of the issue. An engineer can be despatched immediately. In reality, the increase in temperature would have caused SeeWhy to trigger an alert in any case.

The moral is obvious. In our example, no user could possibly set up and maintain individual rules for different times of the day, days of the week for a thousand cell sites and multiple metrics. This needs to be done automatically by the system.  Monitoring needs to happen at very low levels of granularity, and use historic data to provide a baseline for real time data to be compared against.

Of course if you're monitoring individual purchases of 20 million consumers, perhaps from credit card transactions, without this automatic rule capability, any monitoring rules have to be set at an aggregate level, meaning that the critical detail gets lost. And you miss the opportunity to act.

SeeWhy includes a real time dashboard capable of showing real time events in the context of history. But we don't expect you to have to sit and watch it!